| KEYWORD |
Invisible monitoring: implementation of an FMCW radar imaging algorithm for motion classification
keywords MACHINE LEARNING, RADAR, SIGNAL PROCESSING
Reference persons DANILO DEMARCHI, PAOLO MOTTO ROS
External reference persons Irene Buraioli (irene.buraioli@polito.it)
Research Groups MiNES (Micro&Nano Electronic Systems)
Thesis type EXPERIMENTAL
Description Monitoring of human activities has always been a necessity in many applications. HMI (Human Machine Interface) applications in various industries, systems for monitoring patients at risk in the medical field, or plans for monitoring drivers and pilots are only a few examples. Solutions currently in use involve wearable contact systems, which limit the subject functions, sometimes being uncomfortable, irritating, and therefore less effective. The development of 'invisible' systems overcomes many of these limitations, and miniaturized radar technologies are among the most promising.
This thesis aims to analyze radar data in MIMO (Multiple Input Multiple Output) mode with the consequent implementation of a machine learning algorithm to classify human activities correctly.
The proposed thesis work will be structured as follows:
1) The candidate will be required to perform a state-of-the-art analysis regarding the most widely used algorithms for radar data analysis and corresponding machine learning algorithms for human activity classification.
2) Next, the candidate will have to implement the proposed solutions for radar data analysis, resulting in the acquisition of a dataset by defining a group of activities to be identified.
3) Finally, the candidate should validate the proposed machine learning algorithm on the acquired dataset and evaluate its effectiveness and performance.
This process will allow the candidate to demonstrate effectiveness in classifying human activities with an 'invisible' solution that can be reused in various industrial applications.
Required skills Signal processing, programming in Matlab, basic knowledge of ML
Deadline 14/12/2024
PROPONI LA TUA CANDIDATURA